The Study On Data Acquisition And Reconstruction Algorithm For Electrical Impedance Tomography Systems

Electrical Impedance Tomography(EIT)is a new noninvasive functional imaging method.It has become a promising imaging method in the field of biomedical imaging due to its merits such as non-injury,quick-response,small volume,low cost,continuous monitoring,etc.A few commercial products of EIT has appeared in recent years.However,EIT has not been promoted since the inverse problem of EIT is ill-posed and it has low spatial resolution.Besides,solving the inverse problem will take huge computational cost.In the purpose of improving the imaging resolution of EIT,we will do some research in the following two aspects.(1)Digital Lock-in Amplifier(DLIA).By improving the precision of the measured voltage signal,we can improve the ill-posed of EIT.We will design a DLIA,which is composed of preamplifier,AD conversion circuit,Digital Phase Sensitive Detector(DPSD)and low pass filter,to complete data processing.We focus on the design of narrow-band FIR low pass filter.The narrow-band FIR low pass filter is composed of CIC filter,half band filter and shaping low-pass filter,and it is designed with 2MHz sampling rate,0.1Hz passband frequency,3Hz transition width and 90dB stopband attenuation.The performance of DLIA is simulated and analyzed on the MATLAB platform.The simulation results show that there is a linear relationship between the amplitude of input and output voltage signal with 0.1V?1.5V input signal.The relative error is smaller than 0.3%.The DLIA can filter noise well and extract the amplitude of input signal.(2)Reconstruction algorithm.We can improve the ill-posed property of EIT inverse problem by introduce regularization technique when using Newton iteration algorithm.We use Newton iteration algorithm to reconstruct image in this paper,comparing the reconstructed image quality of total variation regularization and Tikhonov regularization,which is based on different prior information.The results show that total variation regularization locate the target accurately,give better shape and size of the target and has good abilities of keeping edge when reconstruct from a single target image,while it can't distinguish targets effectively when reconstruct from multi-target image.The image reconstructed by Tikhonov regularization based on NOSER prior information has good reconstruction image and better anti-noise performance.It can distinguish targets better than total variation regularization in case of multi-target image.